OpenType, Bengali and reph/vattu

eulex's picture


I'm trying to understand how OpenType and Uniscribe is supposed to work, specifically for the Bengali language, and I've discovered one thing I can't seem to understand. It's about the sequence Ka (09af) + Halanth (09cd) + Ra (09b0).
As the Bengali specification states, this is first reordered into Ka + Ra + Halanth by Uniscribe before it's sent to the OpenType rules, then the blwf feature is supposed to turn Ra + Halanth into the Vrama character (named bn_ra_phala in the Vrinda font), then the vatu feature will turn Ka + Vrama into the Kro character (bn_k_ra in Vrinda).
However, before these two rules are applied, the rphf feature is applied, which turns Ra + Halanth into the Reph character. So, Ka + Ra + Halanth will turn into Ka + Reph and not Kro.
But I see that Windows does manage to do this correctly anyway, so what is the problem in my reasoning?

Regards, Alexander Toresson

eulex's picture

Well, I found out the answer. While preprocessing characters, Uniscribe tags characters with what features should be applied to them. Therefore it applies the blwf feature and the rphf features in the right cases.

Syndicate content Syndicate content